Understanding Mathematics Anxiety - Investigating the experiences of UK primary and secondary school students - Cambridge Repository
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
CENTRE FOR NEUROSCIENCE IN EDUCATION, UNIVERSITY OF CAMBRIDGE Understanding Mathematics Anxiety Investigating the experiences of UK primary and secondary school students Report authors: Emma Carey1 Dr. Amy Devine1,4 Francesca Hill1 Dr. Ann Dowker3 Dr. Ros McLellan2 Dr. Denes Szucs*1 March 2019 Author affiliations: (1) Department of Psychology; University of Cambridge, UK; (2) Faculty of Education; University of Cambridge; UK; (3) Department of Experimental Psychology; University of Oxford; UK; (4) Cambridge Assessment English; Cambridge; UK. * Correspondence: Denes Szucs; ds77@cam.ac.uk
Table of Contents Executive Summary ............................................................................................................................ 2 1. Background ........................................................................................................................ 2 2. Aims ................................................................................................................................... 2 3. Methodology...................................................................................................................... 2 4. Summary of Key Findings ................................................................................................... 3 5. Conclusions ........................................................................................................................ 4 6. Recommendations ............................................................................................................. 4 7. Acknowledgements............................................................................................................ 5 The need for this research .................................................................................................................. 6 8. Maths in the UK ................................................................................................................. 6 9. What is maths anxiety? ...................................................................................................... 6 Aims .................................................................................................................................................... 8 Methods .............................................................................................................................................. 8 Key Findings: In Full .......................................................................................................................... 10 10. Validating the modified Abbreviated Mathematics Anxiety Scale (mAMAS) .................. 10 11. Exploring the relationship between mathematics anxiety and performance ................. 14 12. Relationship between maths anxiety and developmental dyscalculia ............................ 17 13. Maths anxiety: Gender differences, developmental change and anxiety specificity ...... 19 14. Anxiety profiles and their relationship with performance .............................................. 22 15. Working memory, dyscalculia and maths anxiety in Italian students ............................. 26 16. Qualitative research: Experiences and origins of mathematics anxiety .......................... 28 General Conclusions ......................................................................................................................... 52 17. Recommendations ........................................................................................................... 53 Outcomes .......................................................................................................................................... 55 18. Publications ...................................................................................................................... 55 19. Cambridge Science Festival Event.................................................................................... 56 References ........................................................................................................................................ 57 1
Executive Summary Background The project investigated individuals’ attitudes towards mathematics because of what could be referred to as a “mathematics crisis” in the UK. Evidence suggests that functional literacy skills amongst working-age adults are steadily increasing but the proportion of adults with functional maths skills equivalent to a GCSE grade C has dropped from 26% in 2003 to only 22% in 2011 (National Numeracy, 2014). This number is strikingly low compared with the 57% who achieved the equivalent in functional literacy skills (National Numeracy, 2014). While mathematics is often considered a difficult subject, not all mathematics difficulties result from cognitive difficulties. Many children and adults experience feelings of anxiety, apprehension, tension or discomfort when they are confronted by mathematics. This may be contributing to a relatively low level of numeracy amongst UK adults. In this project we investigate primary and secondary school students’ experiences in the mathematics classroom and beyond, in order to learn more about the prevalence, nature and resolution of mathematics anxiety Mathematics anxiety describes feelings of apprehension, tension or discomfort experienced by many individuals when performing mathematics or in a mathematical context (Richardson & Suinn, 1972). It has been associated with cognitive difficulties performing mathematical tasks, potentially because anxiety interferes with our ability to hold and manipulate information in mind (working memory), but is predominantly an emotional problem (Ashcraft & Krause, 2007). Developmental dyscalculia, on the other hand, is a cognitive difficulty in acquiring mathematical skills (Devine, Hill, Carey, & Szűcs, 2017). Aims This project had multiple goals. Firstly, we wished to provide estimates of mathematics anxiety prevalence amongst UK primary and secondary school students. Secondly, we wished to validate an instrument for measuring mathematics anxiety in this group. Thirdly, we wanted to see how mathematics anxiety is related to other individual factors, such as maths performance, gender and individual cognitive differences. Additionally, we used interviews to gain a richer understanding of students’ experiences with and feelings towards mathematics. Methodology Our study had two main phases. In the first of these, we worked with more than 1700 primary and secondary students to screen for mathematics anxiety, test anxiety and general anxiety, and gain a measure of mathematics and reading performance. In the second phase, we worked one-to-one with the children to gain a deeper understanding of their cognitive abilities and feelings towards mathematics, using a series of sessions administering cognitive tasks, questionnaires and interviews (Carey, Devine, et al., 2017; Carey, Hill, Devine, & Szűcs, 2017; Devine et al., 2017). Throughout, we 2
worked on two further sub-projects looking at the relationship between maths anxiety and performance and cross-cultural comparisons (Carey, Hill, Devine, & Szucs, 2015; Hill et al., 2016). Summary of Key Findings o We have validated, in our large sample of British children, that the modified Abbreviated Math Anxiety Scale is reliable (it appears to measure one construct) and valid (this construct seems to really be maths anxiety, rather than another form of anxiety or other feelings towards maths (Carey, Hill, et al., 2017). See Validating the modified Abbreviated Math Anxiety Scale (mAMAS) for more details. o We conducted a literature review into the long-established relationship between maths anxiety and performance (those with higher maths anxiety tend to have poorer maths performance). We conclude that this is likely because anxiety interferes with performance and poorer performance increases anxiety, acting as a vicious circle (Carey et al., 2015). See Exploring the relationship between mathematics anxiety and performance for more details. o In our large sample of British children, we investigated the relationship between maths anxiety and developmental dyscalculia. We found that whilst more dyscalculics than typical children met criteria for maths anxiety, the majority of those with maths anxiety had normal performance (Devine et al., 2017). See The relationship between mathematics anxiety and developmental dyscalculia for more details. o In a separate group of Italian children, we participated in research looking at developmental change, gender differences and specificity of maths anxiety. We found that unlike general anxiety, maths anxiety increases with age. The relationship between maths anxiety and performance becomes more specific with age – in younger, but not older, children, this relationship disappears after accounting for general anxiety. See Maths anxiety: Gender differences, developmental change and anxiety specificity for more details. o We have identified, in our large British sample, anxiety subgroups. These may increase in complexity with age. In our secondary school students, we found that those with anxiety specific to academia (high maths and test anxiety) had poorer performance than those with higher, but less specific, anxiety. We conclude that this may reflect a dual path in anxiety development and maintenance (Carey, Devine, et al., 2017). See Anxiety profiles and their relationship with performance for more details. o In our smaller subsample of British students, with whom we conducted further testing, we looked at the relationship between various cognitive variables and maths performance. It seems that a myriad of factors are associated with maths performance, but that basic numerical processing is not (unpublished data). See Cognitive factors in mathematics performance for more details. o In another Italian sample, we investigated specific memory subtypes and their relationship with maths anxiety and dyscalculia. Whereas maths anxiety appears to be associated with a deficit in verbal working memory and perhaps also visuospatial working memory, dyscalculia is associated with deficits in visuospatial memory; both short-term and working memory are affected (Mammarella, Hill, Devine, Caviola, & Szűcs, 2015). See Working memory, dyscalculia and maths anxiety in Italian students for more details. o Our qualitative research has shown that children of 9-10 years are able to discuss their experiences and origins of mathematics anxiety. Mathematically anxious children seemed to describe negative events with less contextualisation. They were also more likely to discuss 3
physical sensations in their maths classes and clearly articulated some of the negative consequences of maths anxiety. See Qualitative research: Experiences and origins of mathematics anxiety for more details. Conclusions Each of the completed projects within our study further reveals the complex, multifaceted nature of mathematics anxiety. It is likely that mathematics anxiety is not a simple construct with only one cause – rather, it can emerge as a result of multiple predisposing factors including gender, cognitive abilities and general predisposition towards anxiety, rumination or panicking under pressure. This helps to explain why mathematics anxiety is robustly correlated to a small degree with many constructs (e.g. test anxiety, general anxiety and mathematics ability). We have clearly shown that emotional and cognitive mathematics problems dissociate and therefore require different intervention strategies. Our qualitative analysis of structured interviews suggests that children as young as 9 are experts in their own experiences in mathematics and this can be harnessed to further understand the thought processes underlying maths anxiety. This brings us closer to design effective prevention and remediation programs for mathematics anxiety. Recommendations o The 9-item modified Abbreviated Mathematics Anxiety (mAMAS) scale developed by this project proved to be a reliable tool for investigating math anxiety in school context. o Teachers need to be conscious that individuals' maths anxiety likely affects their mathematics performance. o Teachers and parents need to be conscious of the fact that their own mathematics anxiety might influence student mathematics anxiety and that gendered stereotypes about mathematics suitability and ability might drive to some degree the gender gap in maths performance. o Hence, for parents and teachers, tackling their own anxieties and belief systems in mathematics might be the first step to helping their children or students. o With our research showing that maths anxiety is present from a young age and goes through significant developmental change, we suggest focusing further research on how maths anxiety can be best remediated before any strong link with performance begins to emerge. o The qualitative part of our research shows that children are able to verbalise the suffering that mathematics anxiety causes them. Our qualitative research also points to several potential causes of maths anxiety that could be focused upon by further research. o Teacher training should clearly highlight the role of both cognitive and affective factors behind maths learning in schools. o Policy makers should be conscious that emotional blocks can have substantial impact on learning potential. o Emotional and cognitive problems require completely different interventions. 4
Acknowledgements The authors would like to thank all members of the Centre for Neuroscience in Education who have been involved in this project, from its inception to completion. In addition, we would like to thank all of the schools who accommodated our busy testing schedule; this research would not be possible without the goodwill of the schools with which we worked. We would like to thank all participating students for their candour during interviews and willingness to complete cognitive testing with us. The Nuffield Foundation is an endowed charitable trust that aims to improve social wellbeing in the widest sense. It funds research and innovation in education and social policy and also works to build capacity in education, science and social science research. The Nuffield Foundation has funded this project, but the views expressed are those of the authors and not necessarily those of the Foundation. More information is available at www.nuffieldfoundation.org. Additional funding was provided by the James S McDonnell Foundation 21st Century Science Initiative in Understanding Human Cognition - Scholar Award 220020370, but the views expressed are those of the authors and not necessarily those of the Foundation. 5
The need for this research Maths in the UK We investigated individuals’ attitudes towards mathematics because of what could be referred to as a “mathematics crisis” in the UK. Whilst functional literacy skills amongst working-age adults are steadily increasing, the proportion of adults with functional maths skills equivalent to a GCSE grade C has dropped from 26% in 2003 to only 22% in 2011 (National Numeracy, 2014). This number is strikingly low compared with the 57% who achieved the equivalent in functional literacy skills (National Numeracy, 2014). This reduction in maths competence over time could be driven by a change in people’s attitudes towards mathematics. For example, many people mistakenly hold the belief that maths is a skill we are born with, rather than one you can learn. This attitude could lead to demotivation in learning, because people think that however much effort they put into learning mathematics, they are unable to alter their fixed mathematical ability (Churchman, 2013). With 4 in 5 adults in the UK having a low level of numeracy, low mathematical skill is costing billions to both individuals and society at large (National Numeracy, 2014). Many employers report concern about their employees’ abilities to ‘sense-check’ numbers – i.e. to ensure that numbers reported are in a ballpark range of what you would expect – resulting in higher costs (National Numeracy, 2014). Less numerate individuals also earn less than their numerate counterparts (National Numeracy, 2014). This is especially disturbing given that socio-economic background influences a child’s mathematics achievement by around 10-20% (National Numeracy, 2014). Therefore, this issue could contribute to a cycle of poverty and decrease social mobility. What is maths anxiety? Whilst some problems in maths are related to difficulty understanding the material, other problems might stem from emotional difficulties. Mathematics anxiety (also called maths anxiety) is a negative emotional reaction to mathematics, which can interfere with the ability to perform mathematical tasks. Maths anxiety has many different manifestations, including emotional - for example, feelings of apprehension, dislike, tension, worry, frustration or fear, physical - for example, butterflies, racing heart, struggling to catch your breath or behavioural - for example misbehaving in class, avoiding maths assignments, not studying maths beyond the minimum expected level (Hembree, 1990). Maths anxiety has been seen in young school children from around the age of 6 (Beilock, Gunderson, Ramirez, & Levine, 2010; Krinzinger, Kaufmann, & Willmes, 2009; Thomas & Dowker, 2000; Vukovic, Kieffer, Bailey, & Harari, 2013). However, negative attitudes towards mathematics and maths anxiety appear to increase when children reach secondary school age, persisting into post-secondary education and throughout adulthood (Dowker, Sarkar, & Looi, 2016). It is difficult to define the prevalence of maths anxiety, because measures of maths anxiety are continuous (i.e. each individual 6
falls somewhere on a spectrum) with no clear cut-off as to whether an individual is maths-anxious or not (Devine et al., 2017). Maths anxiety affects individual wellbeing – for example, some students will dread their maths lessons or avoid doing their maths homework due to a dislike of experiencing negative emotions (Dowker et al., 2016; Hembree, 1990). There is also a relationship between maths anxiety and maths performance: individuals who have higher levels of maths anxiety tend to do worse in maths tests (Ashcraft & Krause, 2007; Carey et al., 2015; Hembree, 1990; Ma & Xu, 2004). Importantly, those affected by higher levels of maths anxiety may develop other negative attitudes towards mathematics, avoid or drop out of voluntary maths classes, or avoid careers which require quantitative skills (Hembree, 1990). Since quantitative skills apply to such a broad range of careers, this may impose a severe limit on the life choices of somebody with high maths anxiety. Many potential causes have been identified for maths anxiety. These include some environmental factors, such as: o Negative experiences in class, e.g. doing badly in maths due to poor basic skills (Maloney, Ansari, & Fugelsang, 2011); o Teacher characteristics, e.g. women with anxious female maths teachers are more likely to become anxious themselves (Beilock et al., 2010); o Parental gender stereotypes, e.g. parents expecting that their daughter will struggle more in maths than their son (Tomasetto, Alparone, & Cadinu, 2011). Additionally, intellectual factors may be involved in maths anxiety. Children with developmental dyscalculia (a specific deficit in the acquisition of mathematics skills) and other mathematics learning disabilities have an increased risk of experiencing mathematics anxiety (Passolunghi, 2011; Rubinsten & Tannock, 2010). Furthermore, an individual’s other personal characteristics might influence their maths anxiety. For example: o Gender - girls are more likely to experience anxiety about maths (Hembree, 1990); o Self-esteem, with a lower self-esteem contributing to higher levels of maths anxiety (Abbasi, Samadzadeh, & Shahbazzadegan, 2013); o Learning style (Sloan, Daane, & Giesen, 2002); and o Attitude towards maths – those who generally like maths tend to have lower maths anxiety levels than those who dislike maths (Hembree, 1990). Whilst studies have identified each of these things as potentially related to individuals’ levels of maths anxiety, the direction of the relationship between maths anxiety and any of the above factors is unclear. For example, in the case of the relationship between maths anxiety and maths performance, it is unclear whether those with lower maths performance are more likely to become anxious about maths (supported by Ma & Xu, 2004; Maloney et al., 2011; Meece, Wigfield, & Eccles, 7
1990), whether maths anxiety acts to lower maths performance (supported by Ashcraft & Faust, 1994; Ashcraft & Krause, 2007; Morsanyi, Busdraghi, & Primi, 2014) or whether the relationship is reciprocal, with both causal directions being active (Ashcraft, Krause, & Hopko, 2007; Jansen et al., 2013; Luo et al., 2014). Aims The overarching aim of our research was to investigate how maths anxiety impacts upon primary and secondary school students’ wellbeing and learning. This encompassed several key objectives: 1. To assess the prevalence of mathematics anxiety in a sizeable UK child sample; 2. To provide a reliable, valid and easy to use measure of maths anxiety for UK children; 3. To investigate gender differences in mathematics anxiety; 4. To investigate the relation of maths anxiety to performance; 5. To see how maths anxiety relates to other cognitive variables; 6. To use interviews to shed light on the origins, everyday experience of mathematics anxiety and coping strategies for mathematics anxiety. 7. To raise public awareness of maths anxiety and disseminate our findings. Methods We used a mixed methods approach combining quantitative and qualitative data methods. Quantitative research methods involve measuring people’s performance and attitudes using instruments such as tests and questionnaires. Qualitative research aims to provide more of an insight into individual experiences, e.g. allowing participants the chance to tell us their experiences without being constrained to several limited choices on a questionnaire. Using mixed methods is one of the key features of our research as it enables us to gain both an understanding of the numbers (answering questions like: how many people have high levels of maths anxiety and how does maths anxiety relate to other measurable individual differences?) and a rich understanding of individual experiences. Data collection occurred in two main phases. Phase 1 involved screening approximately 1800 primary and secondary school children for mathematics anxiety, in whole classes or year groups. Screening tests included questionnaires assessing mathematics anxiety, and two other related forms of anxiety: test anxiety and general anxiety. During screening, students also completed age- standardised mathematics and reading tests. Participating schools were located in South East England, comprised a mix of urban and rural schools, and covered a wide range of socio-economic backgrounds. Subgroups of children with the highest mathematics anxiety (a “high maths anxiety” group) or average mathematics anxiety (a control group) were selected to take part in phase 2. Phase 2 involved further assessments of maths anxiety and general anxiety, assessment of the children’s cognitive performance (IQ, working memory, executive function), as well as two semi-structured 8
interviews. The interviews probed children’s feelings about mathematics, the origins of these feelings, their experiences of mathematics in the classroom and coping strategies, as well as the children’s perceived mathematics competence. Full details of the samples, methods, instruments, and statistical analyses can be found in our published articles (Carey, Devine, et al., 2017; Carey, Hill, et al., 2017; Devine et al., 2017). Phase 1 Additional Projects o 1800 students (half in year 4, half in year 7). o Conducted a review of o Attending schools across South East England. the literature o Tested in whole-class or whole-year groups for concerning the around 2 hours altogether. relationship between o We measured: mathematics anxiety o Maths anxiety; and mathematics o Test anxiety; performance. o General anxiety; o Investigation of o Maths performance; mathematics anxiety o Reading performance. in a large sample of Italian school Phase 1 Analysis students. o Investigation of the o Validated our maths anxiety questionnaire, which relationship between was an adapted version of an adult maths anxiety mathematics anxiety, questionnaire. dyscalculia and o Looked at how maths anxiety, test anxiety and working memory. general anxiety cluster in individuals, and how this relates to mathematics performance. o Looked at how maths anxiety relates to cognitive difficulties with maths (developmental dyscalculia). Phase 2: Maths Anxiety Sample Phase 2: Random Sample o 120 students (half from year 5, half o 200 students (half from year 5, half from year 8). from year 8). o Worked with students one to one for o Worked with students one to one around 2 hours each (split across 4 for around 1 hour each (split across sessions). 2 sessions). o Half had high maths anxiety; half had o Students were sampled randomly. average maths anxiety. o We measured: o We measured: o IQ (verbal and non-verbal); o Mathematics anxiety; o Verbal and visuospatial o General anxiety; working memory. o IQ (verbal and non-verbal); o Visuospatial short term o Interviews about origins and memory. experiences of maths anxiety. o Executive functions. Phase 2 Analysis o Use interviews to enrich our understanding of the qualitative experience of mathematics anxiety. o Investigate the cognitive variables associated with mathematics anxiety and performance. 9
Key Findings: In Full The following sections describe the details of our investigation. Validating the modified Abbreviated Mathematics Anxiety Scale (mAMAS) Why we needed a new child maths anxiety measure For our project, reliably and accurately measuring maths anxiety in children aged 8-13 years was of vital importance. Many measures exist for measuring maths anxiety in adults (e.g. Hopko, Mahadevan, Bare, & Hunt, 2003; Richardson & Suinn, 1972), and several measures of childhood maths anxiety have also been developed in recent years (e.g. Ramirez, Gunderson, Levine, & Beilock, 2013; Wu, Barth, Amin, Malcarne, & Menon, 2012). Adult measures of maths anxiety are not suitable for use in children because they often refer to advanced mathematics which children have not been exposed to. For example, the Abbreviated Maths Anxiety Scale (AMAS) refers to anxiety elicited by checking the tables in the back of your textbook, something which none of the students in our sample will have had to do in their maths classes (Carey, Hill, et al., 2017). On the other hand, childhood maths anxiety scales are also limited. Sometimes this is either because they have not been adequately tested for reliability (an assessment of whether a test measures just one construct consistently) or validity (an assessment of whether a test measures the construct which you want it to). Additionally, some childhood maths anxiety scales are only suitable for use with a very narrow range of children. For example, the Child Mathematics Anxiety Questionnaire is only suitable for use during a 2 year period of primary education, because it asks children questions such as “How would you feel if you were given this problem? There are 13 ducks in the water. There are 6 ducks in the grass. How many ducks are there in all?” (Ramirez et al., 2013). This question is likely to elicit much more anxiety in a child younger than the test was designed for (as they may not be familiar with this kind of problem) and much less anxiety in a child older than the test was designed for (as they will find the problem simple, and research shows that more complex problems elicit higher levels of mathematics anxiety). Beyond the issue of age, it is possible that even within a narrow age range, referring to specific problems might not be ideal. For example, it may confuse the relationship between mathematics anxiety and maths performance even further: lower performing children are more likely than high performers to feel anxiety about a specific question which is above their mathematical capabilities, thus it may appear that there is a stronger relationship with maths performance and maths anxiety than is really present (Carey, Hill, et al., 2017). Other tests of maths anxiety in children are the Scale for Early Mathematics Anxiety (Wu et al., 2012), which also refers to specific questions and therefore is subject to the same problem as the Child Mathematics Anxiety Questionnaire, and the Mathematics Attitude Questionnaire (Thomas & Dowker, 2000). The latter does not refer to specific questions and therefore might be more suitable to assess maths anxiety across a wide range of age groups and abilities. However, it has not been subject to intensive testing of its reliability or validity across a range of ages. Furthermore, this scale does not show the relation between maths performance and maths anxiety which is almost 10
universally observed using other measures (Thomas & Dowker, 2000). This might indicate that it is not a valid measure of maths anxiety (Carey, Hill, et al., 2017). Creation of the modified AMAS Because of these problems with child maths anxiety scales, we decided to modify one of the most commonly used adult maths anxiety scales, the AMAS, to be used with British children and adolescents across our sample’s entire age range. The AMAS has consistently been shown to be valid and reliable, despite its short (9 item) length. This includes several validations of translations of the AMAS, suggesting that it might be robust to minor linguistic adaptations. Below is a table showing items from the original and modified AMAS. Some of the original AMAS items measure anxiety about maths tests (those marked with an asterisk). Others measure anxiety about learning maths (Hopko et al., 2003). Our adaptations translated US-English to UK-English and ensured that items would be age-appropriate for children across our sample. The scale has been used previously but not validated (Zirk-Sadowski, Lamptey, Devine, Haggard, & Szucs, 2014). Item Original AMAS Modified AMAS 1 Having to use the tables in the back of a math Having to complete a worksheet by yourself book 2 Thinking about an upcoming math test 1 day Thinking about a maths test the day before before* you take it 3 Watching the teacher work an algebraic Watching the teacher work out a maths equation on the blackboard problem on the board 4 Taking an examination in a math course* Taking a maths test 5 Being given a homework assignment of many Being given maths homework with lots of difficult problems that is due the next class difficult questions that you have to hand in the meeting* next day 6 Listening to a lecture in math class Listening to the teacher talk for a long time in maths 7 Listening to another student explain a math Listening to another child in your class explain formula a maths problem 8 Being given a “pop” quiz in math class* Finding out that you are going to have a surprise maths quiz when you start your maths lesson 9 Starting a new chapter in a math book Starting a new topic in maths Table 1 Items from the original and modified AMAS questionnaires Assessment of the modified AMAS First, we wished to investigate whether the modified AMAS is a reliable test. In order to do this, we looked at whether scores on each item of the scale were related to scores on each other item of the scale. We found that there was a very strong relationship between all items on the questionnaire, suggesting that it reliably measures one construct. Secondly, we looked at the structure of the modified AMAS, to see if it was the same as the original AMAS. In the original AMAS, items measuring Learning maths anxiety were more related to one another than they were to those measuring Evaluation maths anxiety, and vice versa. Thus it can be considered that the original 11
AMAS consists of two subscales, measuring related but distinct forms of maths anxiety. The first subscale (items 1, 3, 6, 7 and 9) measures anxiety about learning maths. The second subscale (items 2, 4, 5 and 8) measures anxiety about evaluation in maths (Carey, Hill, et al., 2017). We used confirmatory factor analysis, which investigates how items from a questionnaire measure “cluster” (i.e. which items are most interrelated to one another) in order to confirm that a measure has a particular structure. This analysis showed us that the modified AMAS had the same underlying structure as the original AMAS. That is, the modifications we made to the AMAS items did not appear to change what was being measured by each item or how each item relates to each other item. The structure of the mAMAS can be seen in Figure 1 (Carey, Hill, et al., 2017). Figure 1 Path diagram showing how items from the mAMAS cluster onto two subscales, measuring Learning mathematics anxiety and Evaluation mathematics anxiety Whilst this was of great interest, it only confirms that the modified AMAS measures one construct which can be subdivided into two sub-constructs. We were also concerned in showing that the broad construct measured by the modified AMAS really was maths anxiety rather than something else. Researchers in the past, for example, have suggested that maths anxiety might merely be one 12
form of test anxiety (anxiety experienced in evaluative settings; Hembree, 1990). It is also possible that, when working with children, the measure could merely assess how anxious the children felt in general about events, behaviours and competence (general anxiety), rather than being specific to emotions regarding mathematics (Carey, Hill, et al., 2017). In order to assess whether the modified AMAS measured something specific to maths, we looked at the relationship between results on the modified AMAS and results on the two other anxiety measures we used, one of which looked at test anxiety and the other of which looked at general anxiety. These measures – the Child Test Anxiety Scale (Wren & Benson, 2004) and the Revised Children’s Manifest Anxiety Scale II (Reynolds & Richmond, 2012) – have been previously validated. In order to see whether the modified AMAS measures something distinct from test and general anxiety, we looked at how items from all three scales formed into clusters which were more related to other items within the cluster than items outside of the cluster. We found that the scales clustered into items measuring five different factors. One of these factors was best understood as representing maths anxiety. This factor was measured by all items in the modified AMAS (Carey, Hill, et al., 2017). The remaining four factors which came out of this exploratory factor analysis were: test anxiety, physical anxiety, off-task behaviours and social anxiety. Of particular interest was that two items from the modified AMAS (“Thinking about a maths test the day before you take it” and “Taking a maths test”) loaded more strongly onto the test anxiety factor than the maths anxiety factor. This might suggest that whilst all items in the modified AMAS do measure maths anxiety (as they all loaded onto the maths anxiety factor), some of them capture a mixture of maths and test anxiety (Carey, Hill, et al., 2017). Therefore, in this study we have shown three important properties of the modified AMAS: 1. It is reliable – each item in the modified AMAS is strongly related to each other item, suggesting that it reliably measures one construct. 2. It has good construct validity – its underlying structure mirrors that of the original AMAS, suggesting that both questionnaires measure maths anxiety. 3. It has divergent validity – items in the modified AMAS measure something different from what is measured by other anxiety scales. Conclusions These three properties are very important when assessing a questionnaire, and suggest that this is a good instrument with which we can measure maths anxiety across a broad age range (at least 8-13 years). This could be of great practical interest to researchers and educational practitioners alike. The short length of the modified AMAS means that it is very quick to administer and score, which is useful to those who have limited time working with each child. Furthermore, having a measure which can be used across a wide variety of ages is of great interest to those involved in assessing many children of different ages, such as educational psychologists and teachers. For researchers it is 13
also very interesting to use a questionnaire with the same structure as one of the most common adult maths anxiety measures, as it may mean that we are able to compare results obtained from child samples to those which we have from adults (Carey, Hill, et al., 2017). Reference Carey, E., Hill, F., Devine, A. and Szűcs, D. (2017). The Modified Abbreviated Math Anxiety Scale: A Valid and Reliable Instrument for Use with Children. Frontiers in Psychology, 8. Exploring the relationship between mathematics anxiety and performance Before investigating the individuals in our own sample, we took an overview of the literature concerning mathematics anxiety and maths performance. Research has long seen a relationship between maths anxiety and performance in maths tests (Ashcraft & Krause, 2007; Devine, Fawcett, Szűcs, & Dowker, 2012; Jansen et al., 2013; Zakaria, Zain, Ahmad, & Erlina, 2012). This relationship is similar to that seen between other forms of anxiety (e.g. test anxiety) and test performance, and consists of a small negative correlation (Mandler & Sarason, 1952). That is, as maths anxiety levels increase, maths test performance somewhat decreases. Whilst this relationship is consistently observed, there is a lack of consistency in the proposed direction of the relationship: that is, does maths anxiety cause people’s performance in maths to decrease, or does having poorer performance in maths cause people to become more anxious about the subject? The possible directions of the relationship between anxiety and performance can be summarised with two different theories: The Deficit Theory and the Deleterious Anxiety Model (Carey et al., 2015). The Deficit Theory The Deficit Theory suggests that people who start out with poorer maths performance are more likely to develop anxiety about maths, as summarised in Figure 2 (Carey et al., 2015). Poor maths performance Increased maths anxiety Figure 2 Image showing a causal relationship from poor mathematics performance to increased mathematics anxiety For example, studies have suggested that children with mathematical learning disabilities such as developmental dyscalculia (which causes reduced maths performance) have higher levels of mathematics anxiety than children without mathematical learning disabilities (Passolunghi, 2011; Rubinsten & Tannock, 2010). Longitudinal studies (studies which follow children over a longer period of their development than most studies, which only provide a snapshot at a specific time point) also suggest that decreased performance in mathematics might be linked to higher maths anxiety in the following school year (Ma & Xu, 2004; Meece et al., 1990). Furthermore, it has been suggested that adults with maths anxiety might have problems with basic numerical processing (number sense), indicating that perhaps their performance was impaired at a very early stage, before they developed maths anxiety (Maloney et al., 2011). Genetic evidence has 14
also been found which indicates that some of the variation in maths anxiety can be explained by genes which affect a person’s maths performance (Wang et al., 2014). However, this research is not definitive. Whilst it might suggest that some individuals develop maths anxiety as a result of poorer maths performance, not all individuals with maths anxiety have any history of a performance deficit. Nor is it the case that all individuals with difficulties in maths go on to develop maths anxiety. In addition, other research suggests that the link between maths anxiety and performance can be driven in the other direction (Carey et al., 2015). The Deleterious Anxiety Model The Deleterious Anxiety Model suggests that the link between maths anxiety and maths performance is driven by anxiety’s devastating consequences on learning and recalling maths skills, as summarised in Figure 3 (Carey et al., 2015). High maths anxiety Decreased maths performance Figure 3 Image showing a causal relationship from high mathematics anxiety to decreased mathematics performance Maths anxiety may have an effect on maths performance at several different levels. Firstly, evidence suggests that people with maths anxiety are less willing to engage with maths tasks at all. For example, people with maths anxiety are less likely to enrol in maths classes (Hembree, 1990), and have a tendency to answer questions quickly but inaccurately (perhaps due to trying to “escape” the anxiety-inducing maths situation; Ashcraft & Faust, 1994). This suggests a tendency towards maths avoidance in those with maths anxiety, which has a negative impact on both learning opportunities and recall in tests (Carey et al., 2015). Secondly, whilst individuals are engaged in maths tasks, maths anxiety might act to distract them from what they are trying to learn or remember. The idea that anxiety could interfere with learning and recall is known as ‘cognitive interference’ – anxiety generates distracting thoughts and sensations which affect memory capacity. This idea is supported by evidence suggesting that those with higher maths anxiety have poorer working memory (memory used to store, process and manipulate information), and that those with maths anxiety do especially poorly in questions which require a high level of working memory to solve (Ashcraft & Kirk, 2001; Ashcraft & Krause, 2007). The idea is that maths anxiety reduces maths performance both by reducing engagement with maths tasks and by making these maths tasks harder to solve by reducing working memory capacity. There is some evidence that the relationship between maths anxiety and performance does operate in this direction. For example, studies which elevate maths anxiety in specific individuals (for example, by making women conscious of gender stereotypes about women being bad at maths) find that this decreases maths performance (Galdi, Cadinu, & Tomasetto, 2013; Gerstenberg, Imhoff, & Schmitt, 2012; Marx, Monroe, Cole, & Gilbert, 2013; Schmader, 2002; Seitchik, Jamieson, & Harkins, 2012; Spencer, Steele, & Quinn, 1999). Other studies have people do a task aimed to reduce maths anxiety, and have observed an immediate performance increase (Park & Ramirez, 2014). 15
Brain imaging data also suggests that anxiety might interfere with maths performance. For example, there is evidence that when individuals with high levels of maths anxiety outperform those with slightly lower levels, this happens alongside activation of the brain regions associated with cognitive control of emotions. This suggests that those with increased emotional regulation are better able to overcome maths anxiety (Lyons & Beilock, 2012). The Reciprocal Theory The mixture of evidence for each of the two theories suggests that in fact they might both play a part in the relationship between maths anxiety and performance. That is, maths anxiety might cause decreased performance and poorer performance might elicit maths anxiety, as summarised in Figure 4 (Carey et al., 2015). Increased maths anxiety Decreased maths performance Figure 4 Diagram showing a reciprocal relationship between increased maths anxiety and decreased maths performance We believe that a model like this one is best able to account for the mixture of data, which suggests that the relationship between maths anxiety and maths performance operates in both directions. Whilst few studies find explicit proof of a bidirectional relationship, this could be because the two different directions are each best supported by different kinds of study. The effect of maths anxiety on performance seems to most easily be observed in studies which change a person’s maths anxiety (in the short term) and see if this affects their performance. On the other hand, longitudinal (long term) studies find it easier to see the effect which performance has on future maths anxiety. In the future, if single studies are to find evidence that the relationship between maths anxiety and maths performance operates in both directions, it could be useful to carry out “mixed methods” research – that is, research which uses multiple types of methods to investigate this relationship (Carey et al., 2015). This research on the relationship between maths anxiety and maths performance helped to guide our interpretation of our own data. The idea that the relationship between maths performance and maths anxiety is bidirectional suggests two things: 1. Those with cognitive learning difficulties in mathematics, such as developmental dyscalculia, may experience higher levels of maths anxiety. 2. Other factors, such as an individual’s predisposition towards anxiety generally might also act to raise maths anxiety. 16
Reference Carey, E., Hill, F., Devine, A. and Szücs, D. (2016). The Chicken or the Egg? The Direction of the Relationship Between Mathematics Anxiety and Mathematics Performance. Frontiers in Psychology, 6. Relationship between maths anxiety and developmental dyscalculia Mathematics anxiety and developmental dyscalculia We used data from phase 1 of our study to inform us about the relationship between maths anxiety and developmental dyscalculia. Developmental dyscalculia is a difficulty in acquiring mathematical skills, experienced by around 6% of children (Devine, Soltesz, Nobes, Goswami, & Szűcs, 2013). Whereas maths anxiety is an emotional problem, developmental dyscalculia is a cognitive one (Devine et al., 2017). Many different cognitive abilities have been linked to developmental dyscalculia. For example, researchers have suggested that developmental dyscalculia could be linked to problems representing magnitude of numbers, poor working memory (the ability to hold information in mind whilst performing another task), difficulties with inhibition, poor spatial skills or problems organising sounds (Devine et al., 2013). It is unclear which of these factors truly underlies developmental dyscalculia, or if dyscalculia in fact represents a diverse range of mathematics learning problems which have a variety of different factors at their root. However, as we found in our review of the relationship between maths anxiety and maths performance, some studies have found that children with cognitive maths problems, such as developmental dyscalculia, are more likely to experience maths anxiety (Passolunghi, 2011; Rubinsten & Tannock, 2010). We wanted to look at the prevalence of developmental dyscalculia and maths anxiety across our large sample of primary and secondary students. We also wanted to look at how gender differences differentially affect cognitive and emotional maths learning problems. Gender differences in mathematics learning problems The gender ratios reported in past studies of developmental dyscalculia are mixed. Some studies show that developmental dyscalculia is present more in boys than girls (Barahmand, 2008; Reigosa- Crespo et al., 2012; von Aster, 2000), others suggest that more girls than boys have dyscalculia (Dirks, Spyer, van Lieshout, & de Sonneville, 2008; Hein, Bzufka, & Neumärker, 2000; Lambert & Spinath, 2014; Landerl & Moll, 2010). Still further studies show no gender difference in dyscalculia prevalence (Devine et al., 2013; Lewis, Hitch, & Walker, 1994). Gender differences in maths anxiety studies are more consistent, with girls showing higher levels of maths anxiety than boys in the vast majority of studies which found a gender difference (e.g. Devine et al., 2012). Gender differences in maths anxiety are shown more in studies of adults and secondary-school level children, indicating that the prevalence of maths anxiety in primary students is less gender-dependent (Hill et al., 2016). Whilst there may not be any more girls than boys with cognitive difficulties in maths, they appear to be more susceptible to emotional maths learning problems (Devine et al., 2017). 17
Why we looked at maths anxiety and dyscalculia As discussed previously, researchers typically observe moderate negative correlations between maths anxiety and maths performance, suggesting that those with slightly poorer performance in maths are more likely to experience maths anxiety (Carey, Hill, Devine, & Szucs, 2016; Hembree, 1990). Various studies suggest that this relationship also holds in individuals with developmental dyscalculia: these people are more likely to experience maths anxiety than their non-dyscalculic counterparts (Passolunghi, 2011; Rubinsten & Tannock, 2010). However, whilst studies have shown an increase in maths anxiety levels in dyscalculic individuals, no research has investigated the prevalence of co-occurrence of maths anxiety and developmental dyscalculia (Devine et al., 2017). Therefore, we decided to investigate the relationship between maths anxiety and developmental dyscalculia in our large sample. First, we looked at the relationship between maths anxiety and maths performance, not only in the whole sample but also in a subgroup who we identified as having developmental dyscalculia. Secondly, we investigated the proportion of individuals who met criteria for both maths anxiety and developmental dyscalculia, and whether there was a different frequency of co-occurrence in girls and boys (Devine et al., 2017). The relationship between maths anxiety and dyscalculia Most people in our sample had low scores in the mAMAS (our maths anxiety questionnaire). As mAMAS scores increase, the number of children who reported that maths anxiety level decreases. We decided to define “high maths anxiety” as having a mAMAS score above the 90th percentile – meaning that individuals were defined as having high maths anxiety if they were in the top 10% of maths anxiety scores. This corresponded to a score of 30 on the mAMAS, meaning students with high maths anxiety were, on average, selecting above “Moderate amount of anxiety” for each scale item. We defined dyscalculia as a maths performance significantly below average (below 85, where 100 represents average performance) with a reading performance at or above the average range (above 85, on the same scale). This way we found the students in our sample who had a selective weakness in maths – i.e. they performed reasonably well in reading but not in maths (Devine et al., 2017). Whilst there was a negative relationship between maths anxiety and maths performance in the sample as a whole, this relationship was not found in the dyscalculic children. This might just be because their performance scores did not vary enough for the relationship to be seen. On the other hand, it might indicate that once your maths performance is below a certain level, it doesn’t really matter how far below that level you are; you are equally likely to experience maths anxiety regardless. We found that the likelihood of experiencing high maths anxiety was 22% for the dyscalculic children, compared with 10% in the sample as a whole. This confers a statistically significantly increased risk of developing maths anxiety in children with dyscalculia than those without (Devine et al., 2017). However, whilst maths anxiety risk is much higher in those with dyscalculia than those without, dyscalculic children still make up an overall small proportion of all of those with high maths anxiety. 18
In those with high maths anxiety, only 11% fell into the developmental dyscalculia group and 12% had global low performance (below 85 in both maths and reading). The vast majority (77%) of children with high maths anxiety had typical or above-typical maths performance. Gender appeared to play a significant role in the development of maths anxiety in dyscalculic children. Whereas boys with dyscalculia had a low chance of having co-occurring high maths anxiety, girls with dyscalculia were at a much greater risk of having elevated maths anxiety. This fits with what we already knew: girls are more susceptible to maths anxiety than boys. This research shows a large degree of dissociation between cognitive and emotional maths learning problems, suggesting that each may require a different intervention (Devine et al., 2017). Reference Devine, A., Hill, F., Carey, E., & Szűcs, D. (2017). Cognitive and Emotional Math Problems Largely Dissociate: Prevalence of Developmental Dyscalculia and Mathematics Anxiety. Journal of Educational Psychology. https://doi.org/10.1037/edu0000222 Maths anxiety: Gender differences, developmental change and anxiety specificity In collaboration with Italian researchers, we investigated gender differences, developmental change and specificity of maths anxiety in Italian primary and secondary school students. This enables us to investigate whether any findings are specific to the UK, or if they are more general. Previous findings Previous research in this area has shown that girls tend to have higher levels of anxiety about maths than boys, despite generally showing similar levels of maths performance (Devine et al., 2012). This may result from a higher disposition towards anxiety in general, as girls also tend to show higher scores on other forms of anxiety (Lewinsohn, Gotlib, Lewinsohn, Seeley, & Allen, 1998). Alternatively, it could relate to gendered attitudes held about mathematics competence and suitability (Tomasetto et al., 2011). Findings related to developmental changes in anxiety are less clear-cut. It is possible that the link between mathematics anxiety and performance does not emerge until later in school (Dowker, 2005; Thomas & Dowker, 2000). Additionally, whilst maths anxiety is by definition a specific form of anxiety experienced towards mathematics, research has consistently shown individuals with maths anxiety to also show higher levels of test and general anxiety (Punaro & Reeve, 2012; Wang et al., 2014). This calls into question whether maths anxiety is truly a maths-specific anxiety form, or just one manifestation of more generalised anxieties (Hill et al., 2016). In this research we aimed to further explore gender differences in maths anxiety; investigate more about developmental change in the relationship between maths anxiety and maths performance and investigate the specificity of maths anxiety by looking at how it relates to reading performance and general anxiety levels. 19
Participants and measures We worked with around 1000 Italian students, 639 attending primary school with an average age of 9 years, and 342 attending secondary school with an average age of 13 years. Children completed arithmetic and reading comprehension tests, alongside questionnaires assessing mathematics anxiety and general anxiety levels (Hill et al., 2016). Our findings: Gender As can be observed in Figure 5, girls had higher levels of both mathematics and general anxiety. This was the case regardless of whether considering primary school students, secondary school students, or both. This confirms previous research indicating that girls report higher anxiety than boys, both in mathematics and in general. Findings by level of schooling were not significantly different for the AMAS or RCMAS-2, though secondary school students scored higher in the AMAS and primary school students scored higher in the RCMAS-2 (Hill et al., 2016). Figure 5 Average maths anxiety (AMAS) and general anxiety (RCMAS-2) scores in different groups, split by a) gender, b) school level, c) gender within primary students and d) gender within secondary students. Our findings: Anxiety specificity Students’ maths anxiety and general anxiety scores were correlated – this means that as student maths anxiety increases, general anxiety scores also tend to increase. This is graphically displayed in 20
Figure 6 below. This shows that whilst those with higher general anxiety tend to have higher maths anxiety, there is still a lot of variability in maths anxiety reports for any given general anxiety score. That means that whilst these anxiety forms are related, one cannot be entirely explained by the other. This correlation persisted regardless of gender and level of schooling (Hill et al., 2016). Figure 6 Scatter plot of individuals’ general anxiety (RCMAS-2) and maths anxiety (AMAS) scores. Our findings: Developmental changes Researchers have shown a consistent relationship between maths anxiety and maths performance in adolescents and adults. However, some have not found the same relationship in younger children. Our research supports this – we found much stronger relationships between maths anxiety and performance in secondary school boys and girls than in primary school. Whilst primary school girls did show some relationship between maths anxiety and maths performance (higher maths anxiety resulting in slightly poorer performance), primary school boys showed no such relationship. Furthermore, once we took into account levels of general anxiety, there was no remaining relationship between maths anxiety and performance in primary school girls or boys (Hill et al., 2016). 21
You can also read